Your Next IT Strategy
by John Hagel III and John Seely Brown
Reprint r0109g
October 2001
HBR Case Study
Off with His Head?
r0109a
David Champion
HBR at Large
The Leadership Lessons of Mount Everest
r0109b
Michael Useem
Different Voice
Genius at Work:
A Conversation with Mark Morris
r0109c
Harnessing the Science of Persuasion
r0109d
Robert B. Cialdini
Torment Your Customers
(They’ll Love It)
r0109e
Stephen Brown
Radical Change, the Quiet Way
r0109f
Debra E. Meyerson
Your Next IT Strategy
r0109g
John Hagel III and John Seely Brown
HBR Interview
Bernard Arnault of LVMH:
The Perfect Paradox of Star Brands
r0109h
Suzy Wetlaufer
Best Practice
Speeding Up Team Learning
r0109j
Amy Edmondson, Richard Bohmer, and Gary Pisano
Tool Kit
Boost Your Marketing ROI
with Experimental Design
Eric Almquist and Gordon Wyner
r0109k
The age of proprietary information systems
is coming to an end, and the age of
shared services is dawning. Don’t panic.
There’s a pragmatic way to make
the change pay off for your company.
Your Next
IT Strategy
by John Hagel III and John Seely Brown
O
ver the past year, as the hype over e-commerce has
subsided, a new chorus of promises about the potential of
the Internet has been gaining volume. The singers this time
are not dot-coms and their backers but rather the big providers of
computer hardware, software, and services. What they’re promoting,
through a flurry of advertisements, white papers, and sales pitches,
is a whole new approach to corporate information systems. The
approach goes by many different names – Microsoft calls it “.Net,”
Oracle refers to “network services,” IBM touts “Web services,” Sun
talks about an “open network environment” – but at its core is the
assumption that companies will in the future buy their information
technologies as services provided over the Internet rather than owning and maintaining all their own hardware and software.
No doubt, many executives are skeptical. They’ve heard outsized
promises and indecipherable buzzwords before, and they’ve wasted
a lot of time and money on Internet initiatives that went nowhere.
This time, though, there’s an important difference. The technology
providers are not making empty promises: They’re backing up their
words with massive investments to help create the infrastructure
Copyright © 2001 by Harvard Business School Publishing Corporation. All rights reserved.
105
Yo u r N ex t I T S t rat e g y
needed to make the new IT approach work. As these efforts continue, over the next year or two, a steady stream
of new, Internet-based services will come on-line, providing significant cost savings over traditional, internal systems and offering new opportunities for collaboration
among companies. Slowly but surely, all your old assumptions about IT management will be overturned.
In this article, we will provide an executive’s guide to
the new IT strategy. We will explain what the Web services
architecture is, how it differs from traditional IT architecture, and why it will create substantial benefits for companies. We will also lay out a measured, practical plan for
adopting the new architecture – a step-by-step approach
that will pay for itself while mitigating the potential for
organizational disruption. Indeed, we believe that two of
the great advantages of the Web services architecture are
its openness and its modularity. Companies won’t need
to take high-risk, big-bang approaches to its implementation. They can focus initially on opportunities that will
deliver immediate efficiency gains, incorporating new
capabilities as the infrastructure becomes more robust
and stable.
The New Architecture
Until now, companies have viewed their information systems as proprietary. They bought or leased their own
hardware, wrote or licensed their own applications, and
hired big staffs to keep everything up and running. This
approach has worked, but it has not worked well. After
years of piecemeal technology purchases, companies have
inevitably ended up with a mishmash of disparate systems spread throughout different units. Over the last
decade, in efforts to merge these “data silos,” many big
companies have invested large amounts of money – hundreds of millions of dollars, in some cases – in massively
complex enterprise-resource-planning systems, which
offer suites of interlinked applications that draw on unified databases. The ERP systems have certainly solved
some problems, but they’ve been no panacea: Most big
companies still struggle with a hodgepodge of hundreds
of incompatible systems. And ERP systems have also created new problems. Because they’re relatively inflexible,
they tend to lock companies into rigid business processes.
It becomes hard, if not impossible, to adapt quickly to
changes in the marketplace, and strategic restructurings,
through acquisitions, divestitures, and partnerships, become fiendishly difficult to pull off. In effect, the compaJohn Hagel III is the chief strategy officer of 12 Entrepreneuring, an operating company in San Francisco. He can
be reached at [email protected]. John Seely Brown is the
chief innovation officer of 12 Entrepreneuring, where he
developed the perspectives in this article, and also serves as
the chief scientist of Xerox. He can be reached at [email protected].
106
nies that have installed ERP systems have replaced their
fragmented unit silos with more integrated but nonetheless restrictive enterprise silos.
The Web services architecture is completely different.
Constructed on the Internet, it is an open rather than a
proprietary architecture. Instead of building and maintaining unique internal systems, companies can rent the
functionality they need – whether it’s data storage, processing power, or specific applications – from outside service providers. Without getting too technical, the Web
services architecture can be thought of as comprising
three layers of technology, as described in the sidebar
“An Overview of Web Services.” At the foundation are software standards and communication protocols, such as
XML and SOAP, that allow information to be exchanged
easily among different applications. These tools provide
the common languages for Web services, enabling applications to connect freely to other applications and to read
electronic messages from them. The standards dramatically simplify and streamline information management –
you no longer have to write customized code whenever
communication with a new application is needed.
The service grid, the middle layer of the architecture,
builds upon the protocols and standards. Analogous to an
electrical power grid, the service grid provides a set of
shared utilities – from security to third-party auditing to
billing and payment – that makes it possible to carry out
mission-critical business functions and transactions over
the Internet. In addition, the service grid encompasses
a set of utilities, also usually supplied and managed by
third parties, that facilitates the transport of messages
(such as routing and filtering), the identification of available services (such as directories and brokers), and the assurance of reliability and consistency (such as monitoring
and conflict resolution). In short, the service grid plays
two key roles: helping Web services users and providers
find and connect with one another, and creating trusted
environments essential for carrying out mission-critical
business activities. The role of the service grid cannot be
overemphasized: A robust service grid is vital to accelerating and broadening the potential impact of Web services. Without it, Web services will remain relatively marginal to the enterprise.
The top layer of the architecture comprises a diverse
array of application services, from credit card processing
to production scheduling, that automate particular business functions. It is this top layer that, day to day, will be
most visible to you, your employees, your customers, and
your partners. Some application services will be proprietary to a particular company or group of companies,
while others will be shared among all companies. In some
cases, companies may develop their own application services and then choose to sell them on a subscription basis
to other enterprises, creating new and potentially lucrative sources of revenue.
harvard business review
Yo u r N ex t I T S t rat e g y
An Overview of Web Services
payment utilities aggregate charges
The Web services architecture has
Shared utilities provide services
three layers. The most fundamental
that support not only the application
for the use of Web services and ensure
layer consists of software standards
services residing in the top layer but
prompt and full payment.
(such as XML) and communication
also the other utilities within the ser-
protocols (such as SOAP and its likely
vice grid. For example, security utilities
include messaging services to facilitate
extensions) that make it possible for
provide such services as authentica-
reliable and flexible communication
diverse applications and organizations
tion, authorization, and accounting.
among application services as well as
to do business together electronically.
Performance auditing and assessment
orchestration utilities that help compa-
Transport management utilities
utilities assure users of Web services
nies assemble sets of application ser-
through which specialized utilities
that they will obtain agreed-upon lev-
vices from different providers.
provide key services and tools. Four
els of performance and will be com-
types of utilities operate over the
pensated for damages if performance
utilities include service directories,
service grid.
falls below these levels. Billing and
brokers, and common registries
The middle layer is the service grid,
Resource knowledge management
and repositories that describe
available application services
The Web Services Architecture
and determine correct ways of
interacting with them. They
also include specialized services
for converting data from one
Application services
Application
service
format to another.
Application
service
Application
service
Application
service
Service management utilities
ensure reliable provisioning of
Web services. They also manage
Web services
Service grid
sessions and monitor perfor-
Shared utilities
Service management utilities
mance to ascertain conformance
Security, auditing
and assessment
of third-party
performance,
billing and payment
Provisioning, monitoring, ensuring quality
of service, synchronization, conflict resolution
to service-quality specifications
and service-level agreements.
The top layer encompasses
Resource knowledge management utilities
Directories, brokers, registries, repositories,
data transformation
a diverse array of application
services that support day-to-day
business activities and pro-
Transport management utilities
Message queuing, filtering, metering,
monitoring, routing, resource orchestration
cesses – everything from
procurement and supply chain
management to marketing
communications and sales.
Standards and protocols
october 2001
Software standards
Communication protocols
WSDL (Web services description
language)
UDDI (universal description,
discovery, and integration)
XML (extensible markup language)
SOAP (simple object access protocol)
HTTP (hypertext transfer protocol)
TCP/IP (transmission control protocol/
Internet protocol)
107
Yo u r N ex t I T S t rat e g y
To illustrate how the architecture works, let’s contrast the way
a typical business activity – loan
processing by a bank – would be
carried out through a traditional
proprietary architecture and the
Web services architecture. Loan
IT departments will need to inteThe shift to the Web services
processing is a complex procedure
grate new sets of skills in such areas
architecture for corporate comrequiring at least six steps (data
as enterprise application architecputing is not only a matter of
gathering about an applicant, valture, enterprise application integraadopting new technology. It will
idation of data, credit scoring, risk
tion, application development, securequire broad organizational and
analysis and pricing, underwritrity, and IT operations. CIOs will
managerial changes as well as
ing, and closing) and involving interactions with a number of other
become knowledge brokers, pulling
the development of new kinds
institutions (checking an applitogether expertise from within and
of capabilities. A particularly big
cant’s credit rating, verifying inimpact will be felt in the corporate outside their companies.
vestment and loan balances, and
IT operations and performance
IT department. CIOs will face
so on). With a traditional IT archiwill increasingly depend on the
new challenges and assume
tecture, the process is usually supeffective integration of external
new roles:
ported by one, very complicated
resources, requiring deeper skills
IT departments will need to
application maintained by an inin structuring and managing relamove in two dimensions simultadividual bank; like a Swiss Army
tionships. CIOs will become relationneously:
outsourcing
many
tradiknife, the integrated application
ship managers, coordinating the
tional IT activities (as credible
does a lot of things, but it may not
efforts of an array of companies.
and reliable providers of services
do any of them particularly well.
IT departments will need to take
emerge) while leveraging internal
And since the costs of maintaining
electronic connections with other
the lead in shaping the standards
capabilities to design distinctive
institutions are high, requiring
required for industries and busiWeb services that can be sold to
leased communication lines and
ness communities to operate effecother companies. CIOs will become
expensive software to link differstrategists and entrepreneurs, assess- tively. CIOs will become negotiators,
ent systems, the necessary interacing areas of competitive advantage their leadership styles shifting
tions are often handled manually
and focusing resources on building from command-and-control to
through phone calls and faxes. The
persuade-and-influence.
new IT-based businesses.
process, in sum, is cumbersome,
costly, and prone to errors.
With the Web services architecallowing companies to purchase only the functionality
ture, loan processing becomes much more flexible, autothey need when they need it, the new architecture can
mated, and efficient. Leased lines are replaced with the Insubstantially reduce investments in IT assets. And by shiftternet, and open standards and protocols take the place of
ing responsibility for maintaining systems to outside
proprietary technologies. As a result, the bank can conproviders, it reduces the need for hiring numerous IT spenect automatically with the most appropriate institution
cialists, which itself has become a significant challenge for
for each transaction, speeding up the entire process and
many companies. Using Web services also reduces the risk
reducing the need for manual work. And rather than
that companies will end up using obsolete technologies;
maintain its own integrated loan-processing system, the
third-party utilities and application providers will be rebank can take a modular approach, using specialized Web
quired to offer the most up-to-date technologies in order
services supplied by an array of providers. It can also shift
to compete. Companies will no longer find themselves
easily among providers, using one service, say, for risk
stuck with outdated or mediocre applications and hardanalysis of loans to restaurants and another for risk analyware. The standardized, plug-and-play nature of such an
sis of loans to hospitals. In other words, the bank will alarchitecture will also make it much easier for companies
ways be able to use the best tool for the job at hand; it will
to outsource activities and processes whenever it makes
no longer have to compromise on performance to avoid
economic sense. (See the sidebar “Big Changes for Your IT
the complexity of integrating proprietary applications.
Department.”)
Clearly, the Web services architecture offers important
Second, and perhaps more important, the Web services
advantages over its predecessor. First, it represents a much
architecture supports more flexible collaboration, both
more efficient way to manage information technology. By
Big Changes for
Your IT Department
108
harvard business review
Yo u r N ex t I T S t rat e g y
among a company’s own units and between a company
and its business partners. When traditional information
systems need to talk to each other, they do so through
dedicated, point-to-point connections. For example, when
a sales-force-management application needs to send information on closed sales to a payroll processing application for the computation of commissions, a programmer
has to write a special piece of code–a connector–to allow
the two systems to communicate. The problem with such
point-to-point connections is that they are fixed and inflexible and, as they proliferate, become nightmares to
manage. With the Web services architecture, tight couplings will be replaced with loose couplings. Because
everyone will share the same standards for data description and connection protocols, applications will be able to
talk freely with other applications, without costly reprogramming. This will make it much easier for companies
to shift operations and partnerships in response to market or competitive stimuli. The loose-coupling approach
of Web services also makes it an attractive option within
an organization. CIOs can use the Web services architecture to more flexibly integrate the extraordinarily diverse
set of applications and databases residing within most enterprises while at the same time making these resources
available to business partners.
Until now, what’s been called e-business has for the
most part been a primitive patchwork of old technologies. Most companies that do business on the Internet
have had to yoke together existing systems with new ones
to create the illusion of integration. A visitor using a corporate Web site may think it’s a single, streamlined system, but behind the scenes, people are often manually
taking information from one application and entering it
into another. Such swivel chair networks, as they have
come to be known, are inefficient, slow, and mistake ridden. Merrill Lynch, like almost all large companies, has
struggled to patch together hundreds of different applications to support its sites for customers. John McKinley,
the company’s CTO, draws an analogy to the Potemkin
villages in czarist Russia, where brightly painted facades
hid the unseemly reality of run-down homes. The Web
services architecture promises to solve this problem. Taking the people out of the network, the architecture will
enable connections between applications – both within
and across enterprises – to be managed automatically.
First Steps to Success
The construction of the Web services architecture is still
in its early stages, and years of investment and refinement
will be required before a mature, stable architecture is in
place. This does not mean, however, that companies
should wait to begin the transition to a new IT strategy;
even today, benefits can be gained by moving to a Web
services model for certain activities and processes. But it
does mean that companies should take a pragmatic, measured approach. Fortunately, the Web services architecture is ideally suited to such an approach: Because it’s
based on open standards and it leverages the capabilities
of third parties, companies don’t have to place big bets at
the outset. They can carefully stage their investments,
learning important lessons along the way. (See the sidebar “Five Questions You Need to Ask.”)
Merrill Lynch’s McKinley, for example, is currently leading a number of initiatives designed to take advantage of
Web services. One initiative is the creation of an innovative portfolio-analysis system for use by brokers and selected customers. By using XML to link disparate systems
Five Questions You Need to Ask
Senior managers need to ensure that
understanding of the areas of high-
zation that may hinder us from
their organizations’ executives are
est business impact?
exploiting the full value of the new
IT architecture, and do we have
thinking ahead about the implica-
Are we moving fast enough
tions of the Web services architec-
today to build our expertise and
initiatives under way to overcome
ture. Here are five questions you can
exploit immediate opportunities
these obstacles?
use to spur your people:
for streamlining intercompany
Are we exerting sufficient leader-
processes, outsourcing activities
ship in shaping both the functional-
a shared vision of the long-term (five
in which we don’t have distinctive
ity offered by providers of Web
to ten years out) business implica-
capabilities and designing Web
services (defining, for example, the
tions of the new IT architecture?
services that we can market to
performance levels required for
other companies?
mission-critical applications) and
Does our management team have
Do we have a transition plan that
balances the state of the architecture’s development with a clear
october 2001
Do we have a clear understanding
of the obstacles within our organi-
the standards needed to collaborate
with our partners?
109
Yo u r N ex t I T S t rat e g y
within Merrill Lynch as well as to integrate information
new applications. Given these constraints, says Hogan,
from partner organizations, the new system will tie to“traditional IT architectures simply aren’t up to the task.
gether customer information, product information, and
The Web services architecture provides the only way to
real-time market data in a flexible, low-cost way. It will
rapidly enhance our IT platform.” By applying node engive the company’s brokers up-to-the-second, integrated
ablement to existing applications at GM and at the dealviews of all the information they need to meet a cusers, new processes can be rolled out incrementally with
tomer’s needs at any given moment. Merrill Lynch is also
relatively modest investment.
using a Web services approach to enable brokers and
For the first stage in the transition, GM is focusing on
clients to access information and applications from a
using Web services to enhance its traditional build-towide variety of devices, including computers, PDAs, cell
stock model, providing a broader set of options for dealphones, and conventional phones. Both of these projects
ers and customers. It has provided dealers, for example,
offer immediate business benefits: They provide an imwith a locate-to-order functionality – a Web services–
portant competitive advantage to the company’s salesbased application that quickly finds specific car models
people while delivering added value to customers.
in the inventories of other dealers. GM is also planning
Merrill Lynch’s experience, as well as that of other
to roll out an order-to-delivery application, which will
early adopters such as General Motors and Dell, offers
shorten the lead time between placing a customer orthree guidelines for other companies looking to get a
der and delivering the vehicle. Such interim steps will
head start.
pave the way to offering the ultimate build-to-order
Build on your existing systems. The Web services armodel, which will require the reconfiguration of manuchitecture should initially be viewed as an adjunct to your
facturing operations and a more sophisticated deploycurrent systems. Through a process we call node enablement of Web services.
ment, you can use Web or application servers to connect
The payoff is expected to be enormous. GM’s long-term
your traditional applications, one at a time, to the outside
goal is to cut in half its $25 billion investment in inventory
service grid, turning them, in effect, into nodes on the Inand working capital. Analysts at Goldman Sachs estimate
ternet. Node enablement is often as simple as creating an
that supply chain initiatives using Web services could ulexplicit record of the connection specifications of an aptimately reduce GM’s operating cost per vehicle by more
plication – documenting, in other words, its application
than $1,000. Yet the staged approach to change allows
programming interfaces, or APIs – along with the appliGM to shift its IT architecture slowly, avoiding disrupcation’s name, its Internet location, and procedures for
tion and focusing only on systems that will deliver real
connecting with it. The existing application is left intact
economic paybacks at each stage of deployment. It also
but is “exposed” so that it can be found and accessed by
allows the company to temper the risk involved in movother applications in the Web services architecture. The
ing to a new technology platform, since GM’s efforts are
process of node enablement should be systematic, driven
tied to the evolution of the architecture.
by near-term needs but shaped by a view of longer-term
Start at the edge. In implementing the new architecopportunities.
ture, early adopters are concentrating their initial efforts
General Motors provides a useful example of this proat the edges of their enterprises – on the applications and
cess. Mark Hogan, the president of eGM, a business unit
activities that tie their companies to customers or to other
created by the auto giant to oversee its consumer Internet
companies. Sales and customer support are obvious exinitiatives, is a strong advocate of the Web services aramples of edge activities, as are procurement and supply
chitecture. Like Merrill Lynch,
chain management. Less obvieGM began with fairly convenously, some traditionally interGM’s long-term goal is to cut in half nal functions can be pushed
tional Web sites connecting the
company with customers and
out to the edges as a result of
its $25 billion investment
dealers. Now, however, Hogan
outsourcing. In the electronics
and his team have developed a
industry, for example, many
in inventory and working capital.
road map for using Web serproduction activities are being
vices to move GM to a dramaticontracted to specialized mancally new build-to-order manufacturing and distribution
ufacturing service providers, creating a need to share formodel, which will enable the company to generate added
merly proprietary applications and data.
revenue and use its assets much more efficiently. This iniWhy is there so much focus on the edge? Because that’s
tiative requires the ability to communicate and collabowhere the limitations of existing IT architectures are most
rate electronically with more than 8,000 dealers, all with
apparent and onerous. Almost by definition, an applicainformation systems of widely differing specifications and
tion on the edge can benefit by being shared. As a result,
sophistication. Few of the dealers have cutting-edge IT
it suffers most from the difficulties in connecting propriskills, and fewer still have the money to invest in major
etary, heterogeneous systems. As GM found, rolling out a
110
harvard business review
Yo u r N ex t I T S t rat e g y
new set of applications to its far-flung dealer network was
next to impossible before the emergence of Web services.
Dell Computer provides a great example of the benefits
of starting at the edge. Dell’s relationships with its suppliers of components and other direct materials are critical to the company’s strategy. The total amount the company spends on direct materials equals as much as 70% of
its revenue, so even modest savings in supply chain costs
Of course, such lean manufacturing approaches often
just push inventory back from the manufacturer to the
supplier. Dell’s goal, however, is to eliminate excess inventory throughout the supply chain. So the company is
now focusing on reducing the buffers held at the hubs.
These stocks could be cut substantially if supply problems could be identified earlier. If, for example, Dell knew
that one supplier was having a problem fulfilling an order
Trying to engage with too many partners too fast is one of the main reasons that so
many on-line market makers have foundered: The transactions they had viewed as simple
and routine actually involved many subtle distinctions in terminology and meaning.
will have a big impact on the bottom line. A related and
equally important concern to Dell is inventory management. In the personal computer industry, where product
prices have recently been declining at 0.6% per week, excess inventory can become very costly.
Recognizing the huge gains possible from more effective supply-chain management, Dell focused its early Web
services initiatives in this area. It began by more closely
connecting its assembly operations with the network of
outside logistics providers that operate the distribution
centers for direct materials – the vendor-managed hubs,
as Dell calls them. Traditionally, the company had to hold
substantial inventory in the supply chain to ensure that
products could be delivered quickly to customers. Its goal
was to fill orders in five days, yet it took suppliers an average of 45 days to fill materials orders. To ensure it did
not run short of key components, suppliers had to maintain ten-day inventory buffers at vendor-managed hubs,
and Dell had to maintain buffers of 26 to 30 hours at its
own assembly plants. In addition, every week, Dell distributed a new 52-week demand forecast to all suppliers.
Today, Dell generates a new manufacturing schedule
for each of its plants every two hours, reflecting actual
orders received, and publishes these schedules as a Web
service via its extranet. Because the schedules are in
XML format, they can be fed directly into the disparate
inventory-management systems maintained by all the
vendor-managed hubs. The hubs always know Dell’s precise materials requirements and can deliver the materials
to a specific loading dock at a specific building, from
which they are fed immediately into an assembly line.
With this new approach, Dell has been able to cut the inventory buffers at its plants to just three to five hours. Explains Eric Michlowitz, the company’s director of supply
chain e-business solutions, “We’ve been able to remove
the stock rooms from the assembly plant, because we now
pull in only materials specifically tied to customer orders.
This has enabled us to add more production lines, increasing our factory utilization by one-third.”
october 2001
for a particular part, it might be able to temporarily remove from its Web store the computer model that used
the part. This would, in turn, enable a reduction in the
stocks of the part held at the hubs. To establish such an
early warning system for its supply chain, Dell is rolling
out an “event management” Web service, again using its
extranet. This service automatically sends out queries on
the status of orders to suppliers, whose own systems automatically send back responses. Dell expects that this
system will reduce hub inventories by as much as 40%
while at the same time significantly improving gross margins by better matching demand and supply.
Create a shared terminology. The move to a shared
IT architecture raises an obvious question of control: Who
calls the shots? Within a single company, a CIO can impose a set of standards governing information technology
(requiring, for example, that accounts always be represented in applications as “ACCTS”). But once a group of
companies, each with different internal systems and standards, begins to collaborate electronically, establishing
clear lines of authority becomes difficult. In some cases,
one company will have the market power to impose standards on its partners, but these situations are rare and,
given the increasing complexity and fluidity of corporate
partnerships, usually unwise. Instead, shared meaning,
and the trust it engenders, must develop much more organically among participants.
Incremental implementation of Web services can aid
this process. By starting with a few long-standing business
partners – as GM did with its dealers and Dell did with its
logistics providers – companies gain room to experiment;
they can establish through trial and error a common technical language. Then, as they learn what works and what
doesn’t, they can expand the orbit of their partnerships
to encompass new companies. Trying to engage with
too many partners too fast is one of the main reasons that
so many on-line market makers have foundered: The
transactions they had viewed as simple and routine actually involved many subtle distinctions in terminology and
111
Yo u r N ex t I T S t rat e g y
meaning. That doesn’t mean that shared standards can’t
be established among large groups; it just can’t be done
easily or overnight. Traditional distributors spend years
learning the shades of meaning used by different buyers
and sellers. A produce distributor, for example, has to
build an understanding of how each of its suppliers quantifies the ripeness of an orange as well as how each buyer
evaluates ripeness. It is only then that the distributor will
have the knowledge and the authority to create a standard rating system for oranges and promote its adoption
throughout the community of buyers and sellers.
XML can be a powerful tool for building shared meaning in Web-based communities, but it’s important to understand that XML isn’t a cure-all. While XML establishes
a common grammar – a framework for sharing meaning – it establishes only very limited semantics. The precise meanings of XML terms still need to be determined
by the actual partners. For instance, a particular XML tag
may refer to the price of a product, but that doesn’t tell
you if it’s the net price after discounts, if it includes shipping, and so on. Subtleties of meaning have to be hashed
out before business can be conducted in all its inevitable
complexity. And don’t expect the meanings, once established, to stay fixed. They will evolve as partners gain
experience and discover shortcomings in their shared
processes.
The service grid will play an important role in helping
business communities build shared meaning, since a set of
utilities will be established to facilitate the development
of trading standards. In many cases, the dominant companies within private trading networks will provide these
utilities. In other cases, industry consortia will take the
lead. RosettaNet is an early example of a consortiumdriven utility. It is defining and promoting the adoption of
standard XML formats to describe processes in the supply
chain of the electronics industry, enabling all participants
to use the same terms to describe activities like issuing
Lynch, GM, and Dell play key roles by providing their
business partners with compelling reasons to use Web
services. Over time, as additional resources become accessible, the benefits of adopting this architecture will
become compelling to more and more companies. Newcomers will find it advantageous to adopt meanings already in use in order to tap into existing applications and
utilities.
A Platform for Growth
Although many of the early uses of Web services will
focus on reducing costs, efficiency-driven initiatives are
only the beginning. Ultimately, the greatest beneficiaries
of this new technology will be companies that harness its
power for revenue growth. (See the sidebar “Unbundling
and Rebundling.”)
The new architecture provides, for example, a platform
for companies to offer their core competencies as services
to other companies. Smart businesses, in other words,
won’t just consume Web services; they’ll also sell them.
That’s exactly what Citibank is doing right now. It saw
that one drawback to early on-line exchanges was the inability to handle payments: Participants would use an exchange to reach agreement on the terms of a transaction
but would then have to process payments either manually
or through specialized banking networks. Leveraging its
deep skill in electronic payments, Citibank quickly introduced CitiConnect, an XML-based payment-processing
service that plugs into existing trading applications.
Here’s how it works. A company purchasing supplies
through an Internet exchange platform, such as one offered by Commerce One, registers information about the
authorization levels for specific employees and the corporate bank accounts to be used for payment. When a
purchase is made, the buyer clicks the CitiConnect icon
on the Web site. An XML message containing payment
While XML establishes a common grammar – a framework for sharing meaning –
it establishes only very limited semantics. The precise meanings of XML terms
still need to be determined by the actual partners.
purchase orders. Such utilities might also be provided by
independent businesses that are focused solely on developing XML or other software standards within an industry or across industries.
Shared meaning will naturally increase as the use of
the Web services architecture expands. In the architecture’s current, early stage of development, incentives for
its adoption are limited because relatively few application
services are available and the functionality of the service
grid is limited. In this period, early movers like Merrill
112
instructions is automatically assembled, specifying the
amount involved, the identity of the buyer, the identity
of the supplier, the bank from which to withdraw funds,
the bank to transfer the funds to, and the timing of
payment. The message is then routed, according to predefined rules, to the appropriate specialized settlement
networks.
The benefits for buyers and sellers are compelling: Sellers cut settlement times by 20% to 40%, and both buyers
and sellers reduce settlement costs by 50% to 60%.
harvard business review
Yo u r N ex t I T S t rat e g y
are choosing the latter path because it’s faster and allows them to
dedicate their scarce resources to
acquiring customers.
In supplying these functions,
traditional companies have an imresources will be supplanted by
Two and a half years ago, Marc
portant advantage. When an apthe much more efficient model
Singer and I wrote “Unbundling
plication provider has to choose
of orchestrating resources. A new
the Corporation” (HBR March–
between sourcing a Web service
kind of business organization – a
April 1999). In that article, we defrom a well-known enterprise with
rebundler focusing on the assemscribed how most companies ena strong track record or from a
bly and coordination of business
compass three very different types
small start-up with an uncertain
processes that stretch across entire
of businesses – customer relationfuture, it will likely go with the established company, as Commerce
industries and markets – will likely
ship management, infrastructure
One did with Citibank. As tradimanagement, and product innova- emerge and gain enormous power.
tional companies begin to make
Of course, exploiting these new
tion and commercialization – and
their capabilities available to other
kinds of growth opportunities
how the Internet would facilitate
companies through Web services,
will require much more than just
their unbundling, leading to much
the importance of the service grid
the new technology architecture.
more tightly focused companies.
will become increasingly apparVery different organizational
The rise of the Web services
ent. Web services for functions
capabilities must be developed –
architecture will not only speed
like invoicing, payments, and logisnot just new skills but also new
this unbundling but will spur the
tics are critical to the companies
performance measurement and
growth of the new companies by
that use them. Without the secureward systems and knowledge
letting them mobilize a greater
rity, reliability, and performance
management approaches. Even
range of resources to reach a
auditing that service grid utilities
can provide to their customers,
more fundamentally, managers
broader set of customers. Freed
few enterprises will be willing to
will need to adopt new mind-sets:
from the straitjacket of existing
offer, much less subscribe to, such
enterprise-centric IT architectures, They will need to focus on creatmission-critical services.
ing new opportunities, largely
companies won’t have to acquire
As the service grid matures and
by helping to define and deploy
new assets to grow (a slow and
companies move aggressively to
standards, rather than simply tryoften treacherous process); they
exploit revenue growth opportuing to adapt to rapidly changing
will be able to rent them, as Web
nities created by the Web services
environments.
services, from third parties. The
architecture, a curious dynamic
– John Hagel III
capital-intensive model of owning
will begin to play out. The distinction between users and suppliers
of Web services will fade. Companies will provide Web services to
Citibank turned an existing operational capability into
others in areas in which they have distinctive expertise
a new service line and extended its reach to a broader
while at the same time buying Web services from others
range of customers through its partnerships with appliin areas in which they lack special skills. Over time, the location providers like Commerce One. And Commerce
cation of particular capabilities – whether inside or outOne, for its part, got happier customers while also assoside the walls of any given company–will become less imciating itself with the respected Citibank brand.
portant than the ability to discover and orchestrate
The relationship between Citibank and Commerce One
distinctive capabilities across enterprises in order to deillustrates a broader pattern that will both speed the
liver greater value to customers. In the process, many
adoption of the Web services architecture and open new
companies will find themselves turned inside out, with
growth opportunities for traditional companies. All
their formerly well-guarded core capabilities visible and
providers of inter-enterprise applications–private trading
accessible to all.
exchanges, procurement services, supply-chain manageThe authors thank Dennis Layton-Rodin, Halsey Minor, Mahmoud
ment services, and so on–recognize that they need to add
Falaki, and other colleagues for their contributions to this article.
new functions rapidly to attract and retain customers.
Reprint r0109g
They have two choices: develop the added functionality
To place an order, call 1-800-988-0886.
themselves or source it from specialized providers. Many
Unbundling and Rebundling
october 2001
113